[Marc] Argh! And just when I was ready to do a two clock experiment, the second clock, I can't start it, and it shows the yellow light of distress! Hello and welcome back. Yep, you heard right, my precious atomic clock just failed. Actually, it’s not really mine, it’s on loan from team member and uber collector Marcel. Sorry Marcel, I broke your clock. However, looking at the bright side,
I’m going to try to repair it in this episode, which is probably going to be entertaining. So, if you follow us, you know that we have already played with Marcel’s HP 5061 atomic clocks in the past. We explained how they worked, and fired one up in a 2020 video. Then, in 2022, we brushed up on our quantum mechanics, and fine tuned them using the Nobel-price worthy Zeeman effect.
At which point we ended up with not just one, but two tuned atomic clocks on the bench - and Marcel had two more, although not working at this point. Hmm, did I just say that we have several atomic clocks? What about using them to check Einstein’s theory of relativity for ourselves? Einstein’s relativity makes two predictions that should affect our ultra-precise clocks: that times changes if clocks move at a different speeds relative to each other, or if they experience different amounts of gravity. That said, unless you move at a reasonable fraction of the speed of light, or go to a super heavy planet, these changes are infinitely small. However, in 1971, scientists Hafele and Keating realized that this effect might finally be measurable, thanks to the new HP 5061A clocks, the very ones we have.
In a world’s first experiment, they flew them around the earth in opposite directions, and proved that the clocks had indeed sped up and slowed down in accordance with Einstein’s theory, the first direct proof of time relativity. More recently, amateur Tom Van Baak, a self-described time nut, managed to re-run the experiment with the improved HP 5071A clocks, the modern successors to ours. He took his 3 clocks, and his 3 kids, up Mount Rainier near Seattle, for a couple of days, then compared them to his Hydrogen Maser master clock, that he also happens to have at his home. Did I say he was a proper time nut? Take a look at his HP clock collection. Anyhow, he was able to measure a 20 ns jump in time due to the lesser gravity at higher altitude. Although it was far less than the 273 ns that was obtained in the Hafele and Keating experiment, it was still measurable. It’s going to be even harder to do with our old and tired HP 5061A’s, which by the way are all from eBay, but we thought it’d be fun to give it a shot.
To see if it’s even possible, I need to see how closely the two clocks can track each other, when they are sitting next to each other in the lab. I’d need them to stay stable to within a few ns over several days to have any hope of measuring a relativistic shift. So, that’s what I was setting up to do, when, predictably, one of the clocks failed.
So, no measurement today. Instead, we need to dive into the innards of this very intimidating quantum monster, and see if we can save it. [Marc] I can't start it, and it shows the yellow light of distress. And fortunately, I don't think it's with the tube. Because the tube is behaving properly, I think. Where is the ion pump?
My ion pump current decreased all the way to zero. But I have no multiplier, and no beam signal. So, if there's no beam signal, I can't measure anything of course. I think that's the fault: that's the multiplier. On the other clock, it is at a healthy... whatever... multiplier... healthy 30. At this point, we need a short bout of elevator music to explain what I think has gone wrong. It’s explained in full detail my 2020 video, link in the doodly-doo, but I’ll just summarize what’s important for this video.
All of the atomic magic happens within a sealed vacuum tube, which I hope is still functioning correctly. Inside this tube, an oven heats up Cesium atoms, which evaporate and form a beam that goes through magnets and a microwave cavity. The first set of magnets selects only the atoms which outer electron is in the +1/2 spin quantum state. Meanwhile, a strong microwave signal is fed into the following cavity. Due to quantum mechanics magic, when the signal is exactly at 9.192,631,770 GHz, and only then, the spin of the Cesium atom is changed to the opposite -1/2 quantum state. Another magnet at the output of
the cavity then selects all of the -1/2 quantum state Cesium atoms, and sends them to a detector. The result is that we get an incredibly narrow peak signal out of the tube, when we hit the magic 9.192,631,770 frequency on the nose. This frequency also happens to be how the atomic second is defined. An electronic feedback loop continuously tunes the frequency of the microwave oscillator, to maintain the signal out of the tube at its peak, guaranteeing that the microwave signal is always at the perfect frequency. It is then divided to give the 5 MHz reference frequency output of the instrument. You can therefore easily imagine, that a large part of the instrument is devoted to the generation of the microwave signal, at 9.192 etc... GHz. This is done by starting with a 5 MHz
crystal, and using a combo of frequency synthesis and frequency multiplication, as shown here. It all starts with the 5 MHz oven-stabilized quartz oscillator. The 5 MHz is then fed to a x18 multiplier, generating 90 MHz. The 90 MHz is then sent to a harmonic generator, a piece of RF magic this time, which generates harmonic multiples of the 90 MHz, extending far into the microwave spectrum. The 102nd harmonic is then selected by a resonant cavity, and sent towards the tube. But that only gets us to 9.180 GHz. Close, but not quite. We are still 12.631,770 MHz off. And that’s why you see a digital synthesizer in the diagram, which generates the 12.631,770 MHz
we are missing. This is also fed into the harmonic generator, and, because of more RF magic, it gets added to the 9.180 GHz frequency, resulting in the 9.192,631,770 GHz we needed at the output. All that so say that I think there is a fault in this circuit. Conveniently, HP added a detector at the input of the A4 harmonic generator, that indicates the level of the 90MHz coming out of the A3 multiplier on the front panel, when it is in the aptly named “multiplier” position. And that level reads almost zero. That’s no good. And that’s why I suspect that the fault is with either
the A3 multiplier, or on the input side of the A4 harmonic generator. Which, while not particularly amusing, is a lot better than a fault in the irreparable, and immensely expensive Cesium tube. [Marc] So, I think a fault has happened in the multiplier. Oh joy, oh joy. So, we need to repair that one.
So, fortunately, this instrument is superbly well documented, in the 300-plus page manual, which I have scanned, so everybody has access to it. And, if I look at it, it gives me a hint of where the fault might be. So, there's a whole chain here that starts at 5 MHz to make, eventually, the 9.1 92 GHz to match the line of the cesium. So, it starts at 5, then there is a multiplier up to 90 MHz, and there's a harmonic generator.
Simple enough if you look at the block diagram, but more complicated in reality. So, here's our oscillator - I know it works, because I get 5 MHz - and there's the 90 MHz generator, which is a simple multiplier. But the error comes from module A4, which is the harmonic generator that does the times 102 multiplication, plus it adds another signal from a synthesizer, to do the remaining digits. So, the error comes on that yellow signal, that's the measurement point. And that's A4.
So, our problem could be in this beautiful board, which is the multiplier, with its schematics here. And it's fairly straightforward. It's a times two to do 10 MHz, then time three to do 30 MHz, another time three to do 90 MHz, and a power amplifier. Or, it could be from the next module up, which is the RF multiplier, which has a step recovery diode hidden in here. And that's where our error signal comes, from here. We are not getting enough signal. And that's an indication of several things. It's an indication
of both how much 90 MHz is coming in, but also of the bias condition of the step recovery diode. So, if the step recovery diode is open or short, this won't be correct. So, I hope, it's the 90 MHz that's not right. Because, if not, it says this is not repairable. This is a factory thing, we'd have to steal it from another instrument. all right so here's the beauty and I think yes not it is three assembly multiplier so is this guy okay RF connection on that side and power connections on that side so this is the 5 MHz to 90 mahz and then it gets connected up here to the uh to the step recovery diode harmonic generator which gets into the tube and and there are two input to this thing there is the 90 MHz uh which gets multiplied by 102 and there's the 12 whatever megahertz which Sy synthesized oh G1 is this cable here oh indeed it comes from the side of the multiplier so it's this fellow that we somehow have to extract then connect to something else all right I got it okay so first suspect yeah and that's the output of e3 so I need to find a way to connect myself to it okay there was a little challenge because this is an SMB connector which is not that usual so I had to make a little adapter and hopefully I can get that in and we are going to hook it up to my param meter and we should get 25 dbm out of that thing so let's plug it in I here the high voltage Supply and then let's turn that on we have nothing we have we went from what 23.7 to 22 it's our multiplier that's not putting enough power out uh let's check that on the scope now that we know that we don't have much power coming out of it safe to get it attached to the scope okay so let's see what that [Music] does all right well we have something trigger trigger trigger trigger 4 M molt Peak to Peak so that's why we have such a low signal and then we want frequency yeah it's 90 MHz so that's the correct thing but it's only 125 MTS and it should be 4vt RMS so there's something wrong with my multiplier so sort of good news bad news bad news we don't have enough 190 MHz but good news it's in the assembly that's easier to debug it's not the RF uh assembly [Music] oh wait oh oh oh oh I see here is our module power on that side and then so 5 MHz in 5 MHz out and 132 Hertz phase modulation which we don't need that's for locking oh I see some metal gaskets in there oh yeah there we go that's our metal gaskets and this look pretty straightforward no similar complexity to our Apollo stuff very linear a lots of tripler and doublers so lots of coils to tune it might be just out of adjustment although that far seems a little bit much soon after so I have hooked up my multiplier module over there I feed it with 5 MHz from this guy which we repairing can't remember in which episode and this has a lot of oomph so I I can get the 3vt peak to Peak which my uh scope generator cannot do and then I give it 17 no 18.7
volts from over here on the terminal and I look at the output and uh if history repeats itself we should see that we have almost no output 1 2 3 and we have plenty of output we have so much I can't measure it so I need to put an attenuator but something is odd it works so I suppose the reason I don't have any output is because I don't have enough input I need to go back in the machine and look at uh what we get out as an input a 5 MHz input uh why I'm at it I I'll retune it it's it's not that difficult and I suppose it's pretty well tuned a few minutes later so it's all tuned uh I put an anator so I could measure it so now I am at 1.47 so I'll have about uh 3 volt peek to peek about twice that because of the attenuator and it was all pretty much in tune except this coil they give you complicated procedure but in a large multiplayer like this it's just easier to tweak the cords until you maximize the signal pretty simple so I gain maybe 30% something like that so this has a clean bill of health uh the fold is somewhere else so we're back in our instrument I hooked myself up to what was going to J2 which should be the 5 mahz and would don't you know it we have tons of signal we have exactly what we should have 1V RMS at 5 mahz and which is about 3 volts Peak to Peak so why didn't we get our output out of the multiplier that is odd maybe pluged in this some plugs weren't working or something let's plug it back in all right so I've reconnected my multiplier and should be on the scope and we have our 90 mahz back but we haven't done anything he just came all by itself maybe there was a connector problem let let's measure how much output we have so this I know what to expect I put a 6B at meter so you got you guys watch this you're going to see before me [Music] and it's 60 dbm plus 6 so I'm at 22 I should be at 25 I am 3db short which I see on the scope too I should be at 4 Vol RMS but I am 2v RMS so it's a little bit short um and I also see that in the uh there is a resistor to select the gain and they went to the the to the minimum of that resistor which maximizes the gain so they have trouble getting to the voltage to before so I don't know if I need to worry about it change a few transistors in the multiplier or just be happy I reced my multiplier to the harmonic generator so we're are now in a functioning condition and I have this on mold so if we turn it on M should have come back it has not but I have 19 mahz so we are War fears have uh become true with something in the harmonic generator it does receive the 1990 mahz there's something wrong with the the SRD module and the problem is that the harmonic generator is wedged in there it's this thing right here it's another piece of RF magic right you come in at 190 MHz you come out at 9 1 GHz I would think maybe the transistor's gone wrong the regulator that would be our hope that's the part they basically tell you it there's nothing for you in there uh send it back to factory uh none of these adjustments are adjustable except this one which is not our problem so we are a bit little bit in trouble okay I think I found a way to do this you can actually disconnect the thing and rotate to tube and how it comes so this has been disconnected Ed and I left the high voltage but there's enough that I I can go in there and hopefully check the transistor that's probably the only thing I can do in that assembly uh and if that if this not the transistor then we have to go back and work on the multipliers right now it's putting half out of what it should U put so that that might be the the root of the problem is nothing is wrong in the multiplier all right same thing is nice gas getting around it so you can see this thing is a uh piece of RF magic the DI is probably in here somewhere the SRD diode step recovery diode then um this is the resonant cavity to pick the one2 harmonic and then there's all kind of uh impis adaptation here there is an attenuator here and two stubs at least three tune stubs in it so uh would be quite difficult to retune so transistor checking um 6 good transistor 6 transistor seems okay nothing the direction okay so that looks like a good transistor I also wiggle the pot in case it's become oxidized that's for the adjustment of bias and then the cap is super hard to get to so I also checked the cap and it's the better than any caps I have I think that's a talum it's it it has all this capacitance and its series resistance is in the hundreds of Milli ohms so it's an excellent cap so I put it all back together and I think it helped a little bit now I have an indication of 12 I had like3 before so whatever wiggling the pot or um moving Connections In and Out helped um if I get it above 20 I am back in the regular operation mode so I'm thinking that if I get the multiplier to double its output I'll get there so let's work on that uh because the multiplier output is definitely half of what it should be so the way this thing works is supposed to be automat automatic gain control and four 4 Vols RMS here and it comes back here is detected and here's the regulator and however there is a factory selected resistor uh that's used to make sure this is at 4 volts and when I measure mine it's all the way to the maximum is this guy over here this guy over there and so the all the way to the maximum gain which is actually the minimum value of this resistor so this is always at Ground I have always the maximum amount of power which is not enough so which tells me they already had problem at manufacturing uh this assembly wasn't producing enough power I am thinking it must be one of the transistors that's a little weak and they are 2 and 708 they're really old weaklings uh transistors I have some of the power transistor to replace them those are 800 MHz transistors here we we can try to replace them see what happens 1 hour later so I've completed my change of transistors I didn't change them all uh I changed this one this one this one these two are original it work worse when I change those two they are must be very sensitive to impedance uh and it gain me a modest amount of power and the one that gave me the most are the two power transistors which actually I uh figured out had been changed before so this one and you cannot see yeah this one over here these were not original Originals were the right type but they were not each they had not HP markings and I have recovered uh 50% more power I don't have exactly two volts I have 1.73 volts remember this is attenuated by factor of two so it's um it's pretty close I I don't I have like 3.5 volts at the output instead of 4 volts RMS and uh so we will consider it good the transist I ended up using now I chose between a bin of I have the exact replacement and I have others and uh those are HP transistor 1854 0535 and supposed to be a little lower uh frequency cut off but they worked way better so and those are the one that ended up there and gave me my Boost in power so hopefully that will make things better so the assembly is back in there and we'll see if we have regain any power it's not super good so nothing that Mi me jump up and down uh let's let's have it warm up it might not be exactly at 5 MHz yeah probably is not because it was cold so it's probably off the tune and it might increase a little longer than a few minutes later yes and as it warmed up it has indeed gone higher so now it's at 20 which is supposed to be the minimum normal it's a lot better than 0.2 0.3 that we had before so let's see if we can lock it so where are we do we have beam I you have a little bit beam ey yeah it's locked [Music] so and it is still cold so that I don't have that much beam current but I have more than before my multiplier is now within range and so I think that's the maximum we can do on this one and we'll see if it allows us to run a longer term stability experiment moments later and look at that our uh beam current is also much higher so we're getting a lot more signal out of the tube this is still at 20 that's the multiplier output that is good I think we uh restored it to where we had it before so I think it's time for round two of experiments with the clocks and see if we can find the long-term stability between two clocks n
2025-01-24 05:06